EP0932816B1 - Method and device for measuring the course of reflective surfaces - Google Patents

Abstract

The invention concerns a method of measuring the course of a reflective surface of an object including the steps of projecting a defined pattern of at least two different light intensities onto the surface to be measured; observing at least one section of the surface by at least one camera; and evaluating the observed section on the basis of the camera data. The invention provides a simply designed and accurately controllable method in that the pattern produces a mirror image in the reflective surface, and in that the observed section includes a section of the mirror image of the pattern. The invention further concerns an instrument for determining the course of the reflective surface of an object, the instrument including a device for generating a light pattern and having at least one camera for observing at least one section of the surface. The invention provides a simply designed and accurately controllable instrument in that the camera is adjusted precisely to a mirror image of the light pattern.

Description

The invention relates to a method according to the generic term of Claim 1. The invention further relates to a device according to the preamble of claim 19.

In practice there are several probing methods for measuring the Known course of surfaces of flat or curved objects, at which, for example, pneumatically extendable pushbuttons against Move the measuring object to the stop and move it Appropriate signal to a computing unit, which the local Distance determined and, based on a variety of such measurements, a Makes a statement about the course of the surface of the object. The stake such measuring method and the associated devices is expensive because with Consideration of the sensitivity of the probe object to be measured can only extend slowly, possibly the object or its Damaged surface (e.g. scratched, dented) and also the Temperature range in which such techniques can be used, strong is restricted. Such tactile measuring methods are therefore non-contact Mostly inferior to measuring methods.

JP-A-61223605 shows an apparatus and a method for Inspect a surface shape using a strip pattern plate a light source is illuminated and the surface of the reflective test object spotlighting. At a drop angle corresponding to the angle of incidence is a camera built up, which takes the captured image to an image processing system emits where it is with one in a processor unit stored pattern of a flat surface is compared. The Recordings are used to assess the flatness of the observed surface, and do not allow quantitative measurement of the surface of objects.

DE-A-24 39 988 shows a device and a method for Detection of localized defects on curved surfaces. Here will highly concentrated light, generated by a light generator, along one Grid move from horizontal and vertical lines, being a multiple the line thickness remains as the distance between the neighboring lines. The Light is projected onto the test object and at one of the projection angle observed various angles using a television camera. The camera observes a section of the image shown, the angle different from laser and camera. A local deformation of the Body part is determined by deviations in the course of the observed line.

The article by Kiyasu et al., "Measurement of the 3-D Shape of Specular Polyhedrons Using an M-Array Coded Light Source ", in" IEEE Transactions on Instrumentation and Measurement "Vol. 44 No. 3 (June 1995), Pages 775-778, shows a measuring stand for measuring polygonal three-dimensional Objects in which the image of a flat lighting device on a reflective surface after one for not reflective surfaces suitable process from a CCD camera is recorded. The evaluation of the image determined for each pixel starting from a normal vector to be calculated, the orientation of a Polygon facet.

The article by Körner, K. et al. "Fast flatness measurement from large-scale objects ", MSR-Magazin, 11-12 / 1995, page 16-18, describes a Optical measuring method for the flatness measurement of thin glasses. Here will starting from the light source, a line grid on a thin glass plate sharp shown, the optical axis making a large angle (84i) to the normal of the Occupies thin glass plate. This figure is under an equally large Angle observed by a CCD line scan camera, from their observation by phase evaluating algorithms provide information about the height or about the Flatness of the thin glass is calculated. Between line grid and thin glass plate a first imaging stage is provided in the optical axis of incidence, and the optical failure axis between thin glass plate and line scan camera is a second Illumination level provided. The first mapping level is the line pattern of the line grid sharply towards the front of the glass, while the second Imaging level again a sharp image of the image on the line scan camera causes. The known technique has several shortcomings: On the one hand, the geometric arrangement of the components is complex because the optical axes from the imaging levels very flat onto the thin glass plate arrive. The light coming from the line grating is on the first imaging level depicted on the glass plate; the line grid itself is opposite optical axis inclined so that the distances of the lines differ fail. With slight changes in the position of the thin glass plate, the straight is measured and therefore cannot be assumed to be constant, the entire facility can be readjusted. At least they are subject to calculated values of a strong fault tolerance. Since the smallest deviations in the The flatness to be measured from an ideally flat surface is already a strong one Distortion of the image of the line grid is the resolution of the Equipment limited. Furthermore, such a device is also pronounced bulky, so that high space costs arise when such a device is used for the online measurement. Especially when retrofitting Such devices can lead to space problems. Finally is to note that the line scan camera is very sensitive to stray light, which makes a costly shielding must be provided; this is because the image is sharply imaged on the camera, with the camera under one Angle to the glass surface is arranged, at which the measuring point otherwise is outside the field of view, so that the camera can also be exposed to other light (and whose fluctuations) can be influenced. For the evaluation of the measurements the imaging conditions must be known and taken into account. Want one is able to carry out a three-dimensional measurement with the known technology this is only possible in such a way that the glass plate is relative to the image section that the Camera is observed, moved, and several measurements in a row occur. For measurements at several locations on the surface at the same time Technology not designed, because only in one area a sharp image of the Line grid is done. As a result, exact measurements take on the order of magnitude one to several minutes. For the measurement of surfaces with larger radii the pattern shown can no longer be evaluated.

US-A-5 110 200 shows a method and an apparatus for Observation or measurement of the human cornea. A video picture a reflection of an illuminated ring is thereby on discontinuities in the Brightness of the image examined for determining the contour of the cornea. However, this procedure is for moving objects, such as moving objects Bands, unsuitable because if there are a lot of changes at the same time the evaluation fails. The procedure for the evaluation is furthermore of objects that reflect light several times, as is the case for the both surfaces of a glass pane is also not suitable. It can also known methods no deviations in the flatness of a surface determine. Finally, the known method is quite time-consuming, so that it is not suitable for industrial use.

DE-A-44 01 541 also shows a method for determining the Surface structure of the cornea. One detects behind a pinhole Arranged camera by a light emitting diode, which on circular orbits with variable Radii circle, emitted and reflected from the retina Light. Deviations of the reflected light from the circular shape show up local defects in the retina closed. Because of the intermittent Observation can also take place in the case of merging or interrupted reflections this the respective output radius of the LED be assigned. The known method is time consuming and therefore not applicable on an industrial scale. Nor is it for double reflective materials suitable. Nor is it possible to move the Light source to overlay a movement of an object to be measured.

It is the object of the invention, a method according to the Preamble of claim 1 or a device according to the preamble of Claim 19 to specify, which are simple and precisely controllable.

This task is performed with the method mentioned at the beginning the characterizing features of claim 1 solved. This task will in the above-mentioned device with the characteristic features of claim 19 solved.

The technique according to the invention in particular allows measurement the course of a reflective surface of an object that is made of a partially transparent at least for light of certain wavelengths Material exists that this light on at least one other behind the Surface arranged surface reflects. Such items are for Example glass plates, plastic films, with a transparent coating provided reflective surfaces, laminated car windows, etc. The superimposed reflections of the at least one are preferably two areas separated, so that measurements of a previously unprecedented Precision arise.

It is both possible for the camera to mirror the image directly observed as well that the camera indirectly through a mirror image Mirror arrangement observed.

In the first case, the optical ones expediently close Axes (planes) between pattern and mirror image on the one hand and between Mirror image and camera on the other hand an angle that is less than 90 °, preferably (very) much smaller. The two optical axes are preferred so arranged that they reflect light like the angle of incidence and angle of reflection Normals of a flat surface have the same angle on both sides of the normal take, so the sum of the included angle. It is possible and preferred if this included angle is very small. The Opening angle between pattern and camera is the angle (at which plan Surface) between those levels in which the line scan camera and the observed mirror image detail on the one hand or the light pattern and the observed section are arranged on the other hand. With a matrix camera the corresponding angles for each line are preferably close together.

In the second case, where a mirror arrangement is provided , the latter preferably comprises a parabolic mirror that observed Mirror image regardless of the distance between reflecting surface and always shows the parabolic mirror sharply at one point, in which the camera is arranged.

An easy-to-implement pattern shows equidistant, alternating bright and dark streaks of light; but also other geometrically defined ones alternating light-dark sequences are suitable as patterns, e.g. such with Stripes that have more than two different light intensities, checkerboard patterns, checkered pattern. A pattern made from dark is particularly advantageous (light) stripes crossing each other at right angles, which Include light (dark) squares, the edge length of which is the width of the stripes equivalent.

According to the invention, the pattern is immediately reflected in the Mirrored surface. Flatness errors (i.e. small gradients) in the measured 2- or 3-dimensional reflective surface cause distortion in the mirror image, with a slight increase or decrease in the surface a wider or a narrower mirror image induced by the incoming Light is bundled or scattered a little more. With the camera it is possible, the change in intensity (or the Course) in the chiaroscuro mirror image, and thus already the smallest Evaluate changes in the degree of brightness, and for example via a Difference formation with an ideal mirror image Conclusions on fluctuations in the surface to be measured. Based on border changes in the dimensions of the light-dark distances observed by the camera the local differences in inclination are calculated. Because of the simple Structure of the device according to the invention or by the compact and direct The method according to the invention allows the surface course to be extremely rapid and reliably record. Starting from the pattern, the light comes in parallel to the measuring surface. The light-dark sequence of the pattern is reflected in the Surface. Assuming a completely flat surface, if the lateral offset to the normal is not too large - one to the light-dark pattern result in exactly proportional reflection. Since its structure (distances etc.) is known, the arrangement according to the invention advantageously enables the calculation of the deviations from an ideal picture is not in comparison to perform a captured image of a normal surface, but in relation to the known dimensions of the line grid, whereby the Accuracy of the evaluation is improved. The degree of distortion one Streak of the observed mirror image is measured, and from this its inclination determined to a flat surface, the inclination can be expressed as an angle is. For each stripe of the pattern, the inclination of the determine the corresponding patch; you string these pieces together (Integration), starting from a point defined (e.g. by a stop) the surface, you also get information about the successive as a result of the inclination of the height increase, the course can "consequences".

A very interesting special case of measuring the course of a Surface is the measurement of a flat surface to determine its flatness. This information is required in a large number of processing companies. The method according to the invention and the device according to the invention are also suitable for this. Basically, the same evaluation can be used; however, a limit value analysis is preferably carried out, in which the Change in the angle of inclination over the route is determined. Then you can too examines an inclined surface for the accuracy of its surface treatment are evaluated without the incline as such being an undesirable inclination becomes. The average helix angle could also be derived from this. The The evaluation described above can be used to examine the Smoothness of all "steady" surfaces are used, e.g. also for Examination of the sphericity of a sphere. Minima, maxima and turning points in the differentiations show areas with "non-smooth" or do not plan gradients.

The pattern can both be inexpensively arranged in series a light source and a physical grid can be realized as also by means of a matrix of a large number of LEDs.

According to a first preferred variant, the Light pattern that is used to make a structure of at least two regularly alternately arranged different light intensities. To one To obtain optimal contrast is according to a first preferred development a line grid of equidistant lines that alternate opaque (light transmission approx. 0%) and transparent (light transmission approx. 100%) uses that from a light source through the transparent line light passing parallel to the reflecting surface with the light passing through the opaque line at the passage through the light grid hindered light one one Sequence of alternating light and dark lines or stripes generated that are reflected in the surface. It is understood that the translucency to light of a certain wavelength or a wavelength range can be limited. But it is also according to an alternative advantageous variant possible to design the light grid so that it in Sequences has more than two different light transmittances, for example 0% -50% -100% or 1% -10% -100%.

Depending on the distance of the camera, which is preferred is a line or matrix camera, the image results for each pixel (Pixel) in the camera is a measurement that is either observed by a dark Line or an observed bright line of the mirror image or in the transition area between two lines is expressed by a gray value. comparing one measured these values with the values of an ideal mapping (and one assumes a parallel course of the light, which in the first approximation without Loss of accuracy is allowed, making the comparison direct with the Light pattern itself can be done), so you easily get locally definable Deviations from a true-to-scale light pattern. Based on these Deviations become the smallest deviations from a flat surface Inclinations can be calculated precisely and in an evaluation unit, for example with Determined with the help of a software routine. An evaluation is advantageously carried out the relative position of the observed light-dark sequence, so that a possible bevels in the arrangement of the otherwise even plan the surface Measurement not affected.

According to a second preferred variant, the Light grid around a cross grid, which essentially looks like a line grid, where, for example, the opaque lines are also perpendicular to the first Line direction. So there is a cross grid that is a quarter is translucent and three-quarters opaque. However, it is an alternative possible, in the light grid the one surrounded by opaque rectangles enlarge the translucent surface in such a way that the ratio light / dark approximately is equal to. Again, it is also possible to have more than two light transmissions provided. A surface camera is used to observe the mirror image used, i.e. a matrix camera, which is a rectangular, two-dimensional Section of the surface observed, thus enabling a 3D measurement (the deviation in flatness calculated from the measured values (Inclination, ripple, height) of the surface corresponds to a first dimension; this value is measured using the coordinates in the longitudinal and transverse directions, so that the surface can be represented as a three-dimensional image; a measurement with the line scan camera would be a 2D measurement: course over length, which e.g. at certain vehicle body panels makes sense).

According to a third variant, it is possible to use the light grid as Form checkerboard structure, in which alternate squares opaque and are designed to be translucent.

According to a fourth variant, those in the previous ones Variants achieved sequences of light intensities through a matrix of LEDs generated that are designed similar to a stadium display in a sports arena can.

The methods described above have in common that they observed reflection against the known dimension of the grid compare, and deduce the angle from observed deviations to which the surface is inclined towards, for example, a flat surface.

A highly precise measurement of the flatness or smoothness and the Waviness of the surface of a reflective surface, its accuracy and Resolution is improved once again by a factor of 10 to 50, can be improved by a Achieve advantageous execution: The reflection of the grid is also observed by a line or matrix camera, but in such a way that in each case for each dimension a light-dark sequence of the pattern, preferably one equidistant light-dark pair, e.g. through an opaque and transparent Light grid as the cross grid mentioned above is generated to a number of Pixels of the camera is mapped, which is an integer multiple of the sequence is. This creates moire beats on the "grid" of the camera, the extremely precise statements about using phase-evaluating methods Allow deviations in flatness. What is recorded is evaluated during the evaluation Moiré image into a sine curve typical of the moiré image (or another cyclic curve) and from the phase shifts on the one hand and the compressions and strains of the calculated sine curve on the other hand inferred from flatness errors, waviness and the like.

The ratio of one light / dark pair to three is particularly preferred, alternatively also to four or five pixels. Because of the high cost of It is useful for matrix cameras to increase the number of pixels per light / dark sequence minimize.

In the case of, for example, a grid with a checkerboard pattern light and dark squares and a matrix camera, this means that on four Squares come nine pixels. The pixels are directed towards the image in such a way that Ideally, one pixel is a full bright and another pixel is a full one dark spot observed, while the pixel in between became a shade of gray detected. If there are changes in the surface polarity, the image will be in Dependence on the fluctuation shifted and the light / dark values that be captured by the pixels by a certain amount in one or the other other direction shifted. This amount is particularly easy to get from Determine intensities that measure the pixels and simple evaluation procedures attributed to an angular displacement that is an expression of the flatness. As a result of moiré images, which are due to the superposition of the Light grid and the "grid" of the line scan camera can create information can be determined using the flatness with a significantly higher resolution.

It is also possible to mirror the reflection of a light / dark sequence map exactly two pixels, which makes the camera, especially one Matrix camera, becomes much cheaper. Then, however, the light grid is to be designed in such a way that sequences of lines are created that are in at least three repeat different light intensities. Also leaves this ratio determine a sine curve, the phase shift of which is formed Moire beat can be used to determine flatness errors.

Finally, it is also possible to use the mirror image of a light-dark sequence, especially a pair, on a pixel or a multiple thereof mapping; but then you need three shots, each of which Pattern is shifted by a third of a light-dark sequence. This shift can be easily realized with the matrix of LEDs already described.

With at least partially transparent materials starting from a separate measurement of the bottom and the Top of glass reflected light beam in addition to analyzes of height and Waviness also draws conclusions about the optical refractive indices of the material take place that are otherwise only possible with transmission methods. That's the way it is possible according to a particular advantage of the invention, from one of a Plurality of superimposed reflections to isolate that reflection, which comes from the back of the material, and where the lighting pattern (or the illuminated parts of the pattern) have passed through the material. This reflection on the "back" falls (e.g. because of a loss of not reflected light) so that it can be easily isolated. The Reflection on the back also provides one after differentiation Statement about the reflection optics of the material.

The measurement with the methods or devices according to the invention is very fast and only takes a few milliseconds on the order of magnitude. It is therefore advantageously possible to also "flow" material, for example at the exit of a rolled glass system or endless rollers made of reflective steel, to watch and measure. When using a cross grid with a It should be noted that in contrast to the line scan camera, which the Reads out pixels in parallel, matrix cameras are usually read out serially. There the time interval between two read processes may be high Conveyor speeds of the surface to be measured is too high possible to illuminate the light grid using a stroboscope or flash shorten. If, on the other hand, the surface is stationary, it can preferably be used with a Line scan across the entire surface, and one from it quasi three-dimensional representation can be determined.

It is preferably taken into account in the evaluation step that the Distance of the line scan camera to the surface to be measured over the line of Camera is not quite even, since the line scan camera has a shorter extension has as the extension of the observed mirror image. Accordingly, the Camera in the direction of the mirror image on an aperture that in is generally very small, but to a certain degree of inaccuracy (as a result Blurring and due to the larger distance) leads to the evaluation of the picture are taken into account and essentially already by the Image ratio on the camera is balanced. The above applies also for both dimensions if a matrix camera is used.

According to a particularly preferred development of the invention it is also possible to measure transparent mirror materials, for example Picture tubes, displays, plane glass, mirrors, curved glass or tempered or Laminated safety glass. Such materials have the property that they reflect incident light on both its top and bottom and thus give a double image. The image of the top of whose Intensity usually is slightly stronger than the image on the bottom, overlaps with the picture of the bottom. While trying in the prior art. by angles that are as flat as possible are to eliminate the image of the underside preferably means are provided that take into account the image of the bottom enable in the evaluation.

According to a first variant of these means, this is implemented in such a way that the image of the image on the camera is out of focus, so that both reflected images come to fruition with their resulting intensity Camera detects a signal that comes from both mirror images. It is to take into account that depending on the thickness of the glass the path of the Light paths are of different lengths and therefore a corresponding offset occurs, which manifests itself in that between a bright area (itself superimposed reflection of the transparent strip) and a dark area (overlapping reflections of the opaque area of the grid) "cloudy" Areas are given where there is the "reflection" of a dark area (More precisely, there is no reflection of darkness, but the corresponding dark stripes is delimited by two reflected light stripes) one side of the glass with the reflection of the bright area of the other Side of the glass overlaid (and vice versa). The distance between these areas depends from the angle at which they are observed; this angle is known and must be taken into account in the evaluation. In doing so, reflection of a bright grid area on the top the intensity of the captured image be slightly larger than in the opposite case.

Alternatively, and preferably, it is possible to have two in a row to provide arranged line or matrix cameras, which preferably observe the same image section, with one camera at a time from mirror image reflected on one side of the transparent material (focus on Raster) is in focus. The individual evaluation can then be made in the knowledge of Thickness of the item can be made.

By providing two cameras, preferably in different heights and / or at different angles to the surface, can also - both with transparent and with opaque surfaces - discriminated whether a change in the number of light-dark sequences observed in a detail on the curvature of the surface (Lens flare) or due to a change in altitude at which the reflective surface is arranged (which makes the proportionality factor between the defined pattern and mirror image).

Basically, it is preferable to double the reflection signals separate reflective materials such as glass. Because besides the more precise Individual information that enables a statement about the "front" surface, also e.g. through an integration step, a statement about the form wins one by differentiating the measured values from the "back" surface through a differentiation step a statement about the Reflective optics at the location of the material being measured.

A particularly useful way of observing the same Clipping is possible if a semi-transparent mirror is provided, which the observation of the same section with different cameras different distances. For example, in one Image detail an increase in the number of light and dark stripes observed, that would be assuming a fixed altitude of the Surface on a curvature in the surface like a converging lens interpret. Assuming a perfectly flat surface, but in their distance from the pattern and camera would fluctuate, the Increase the number of stripes observed in a section to one Increase in the distance to the pattern or camera due to this indicate extended ray paths. The same applies to observed reduction in the number observed. In practice, in particular if the object with the surface to be measured during it Movement is measured, both influencing factors overlap, but it is for the calculation of the course is particularly advantageous if errors due to Fluctuations in altitude can be eliminated. It should be noted here that the cameras for observing the same image section from different Distance have a correspondingly different aperture. That means but that in the event of a deviation in the height of the surface, the number of each of the two cameras observed stripes proportional to the ratio Deviation / distance varies.

Then preferably the evaluation of the two Cameras observed image detail take place so that the deviation among themselves in the number of absolutely observed strips in a first step to Determine the actual altitude used and the number of times at least one of the cameras observed stiffeners in a further step based on the expected number at the determined altitude taking into account the corresponding proportionality factor to determine a Curvature is used.

The inventive evaluation of the surface of a is at least partially reflective object by means of a matrix camera Although cheap in terms of evaluation, matrix cameras are more than certain Orders of magnitude quite expensive to buy. Are essentially plane Measuring materials, the measurement primarily concerns the investigation the flatness, you can get almost the same information as with a matrix camera achieve through a particularly preferred development of the invention, for which only two line scan cameras are required. Here, the reflection of two Light grid observed in the same location on the surface by one camera each. However, the light-dark sequences of the light grid are oblique, preferably 45 ° arranged to the direction of transport of the material, and in one at 90 ° complementary angle, so preferably again 45 °, oblique to it Longitudinal axis. Furthermore, the slopes of the two light grids are accurate opposite, so they preferably cross by about 90 °, are thus inclined to the mirror axis with the camera. If there is an (ideally typical) Change in reflection occurs so that the reflection is in one direction Surface is deflected perpendicular to the longitudinal extent of the grid, so migrates (according to the trigonometic ratio) the image captured by one camera in a direction normal to the direction of deflection by a proportional factor while the other camera emigrates to the opposite to the direction of deflection detected normal direction. In the other (ideally typical) Case in which a compression or an elongation in the reflection in Longitudinal direction of the grid is detected, this change underlying lens effect can be determined in a phase-evaluating manner. There both cameras observe the same section of the material, the discriminate easily between the two (ideal) deviation types and focus on the return the corresponding place of reflection, so that a "mapping" of the Surface is possible. It is understood that in the case of a continuous Scanning the surface with such a device can remain unmoved. This variant according to the invention is particularly suitable for measurements on objects, especially endless under the measuring device be relocated.

Evaluations are also possible in which the Intensities at different wavelengths are taken into account, in particular using a color camera for the red, blue and green intensities or at tinted glass, the light of different color or wavelength different strongly absorbed.

It should be noted that the invention and its developments for the measurement of flat reflective surfaces such as flat glass is also suitable for reflective surfaces, e.g. polished surfaces that have a multidimensional sphere, for example vehicle glass panes, Stamped parts, picture tubes, objects coated with a reflective layer and the like, the objects in addition to rolled, drawn and float glass can consist of acrylic glass or PVC.

Further advantageous developments of the invention result from the following description and from the subclaims.

Fig. 1 ist eine schematische Darstellung eines Ausführungsbeispiels der Erfindung.

Fig. 2 zeigt eine schematische Frontansicht einer erfindungsgemäßen Vorrichtung.

Fig. 3 zeigt eine schematische Seitenansicht der Vorrichtung aus Fig. 2.

Fig. 4 ist eine schematische Darstellung eines Ausschnitts eines von der Kamera erfaßten Signals, das anschließend weiterverarbeitet wird.

Fig. 5 zeigt schematisch in Draufsicht ein weiteres Ausführungsbeispiel einer erfindungsgemäßen Vorrichtung.

Fig. 6 zeigt schematisch in Seitenansicht die Vorrichtung aus Fig. 5.

Fig. 7 zeigt schematisch in perspektivischer Sicht die Vorrichtung der Fig. 5 und 6.

The invention is explained below with reference to an embodiment shown in the accompanying figures.

Fig. 1 is a schematic representation of an embodiment of the invention.

2 shows a schematic front view of a device according to the invention.

FIG. 3 shows a schematic side view of the device from FIG. 2.

Fig. 4 is a schematic representation of a section of a signal detected by the camera, which is subsequently processed.

5 shows a schematic top view of a further exemplary embodiment of a device according to the invention.

FIG. 6 shows a schematic side view of the device from FIG. 5.

FIG. 7 shows a schematic perspective view of the device of FIGS. 5 and 6.

1 to 4, 1 is an arrangement for Measuring the course of reflecting surfaces. The too measuring reflective surface 2 is a 2.5 mm thick curved Rolled glass pane that has a rectangular plan and a bend in the longitudinal direction having. On the Walglas glass 2 is by a light source 3, the extends in a plane above the rolled glass pane 2, projects light that passes through a light grid 4. The light grid 4 consists of equidistant, 5mm wide stripes or lines that alternate between opaque and transparent are trained. This is the parallel passing through the light grid 4 Light arranged in (bright) stripes, which are separated by non-illuminated (dark, Light intensity equal to zero) strips are separated. It should be noted that there the light emerges in parallel from the grid 4, the width of the grid 4 in extends essentially over a width corresponding to the width of the pane 2, around the disc 2 almost completely with illuminated and unlit strips to cover. The length of the grid 4, that is, the extent that extending across the strips and parallel to the short edges of the strips, is approx. 2m, so that a total of 400 strips (200 light / dark pairs) are arranged side by side.

A line camera 5 faces the same surface of the pane 2, the a section of the mirror image 6, which of the grid 4 in the Surface of the glass piece 2 is formed, detected. The neckline is approximately 80cm. The observed mirror image detail, that with the dashed area in FIG. 1 7 is indicated, is located essentially in the center on the surface of the Disk 2 and is substantially perpendicular to the extension of the Light / dark streaks in the mirror image 6. It should be noted that equidistant Strips simplify the evaluation, but are by no means mandatory. The Line camera 5 detects the position and with each of its pixels with high accuracy Intensity of the observed mirror image 6. For this purpose, the focus of the camera 5 set to grid 4.

However, it is possible to have the mirror image 6 in the surface 2 of the Glass directly instead of indirectly as described above, e.g. about Watching mirrors. Preferably comprises a mirror arrangement for the indirect observation for this a parabolic mirror. The parabolic mirror will arranged such that a section of the mirror image 6 on the camera throws back. On the one hand, the image reflected by the parabolic mirror is always sharp adjusted to the camera; on the other hand, this attitude is independent of that Distance of the parabolic mirror to the glass 2, so that the effort of Focusing of the camera is advantageously reduced.

Starting from a completely flat surface, one would the corresponding raster 4 proportional or identical mirror image 6 of the camera 5 can be detected. However, there are differences in flatness on the Surface of the glass 2, there is a distortion of the mirror image 6, that is, the spacing of the strips from one another, which is captured by the camera 5 are no longer equidistant. Deviations from a complete flat surface appear as angle differences in surface 2, it lies A profile of mountains and valleys, so to speak, whose slope is zero (completely flat surface) is different. The mirror image 6 one measured surface 2 is proportional to these gradients distracted. These deviations by which the reflected mirror image 6 falsifies are registered by the camera 5. Since the light grid 4 is very accurate the angle differences in the surface can be exactly and determine extent. Based on these determined angular differences In a planned (target) subject, both the flatness and calculate the waviness of surface 2.

Basically, the same laws apply to one curved surface (or a sphere, etc.). However, the evaluation complicated by the fact that the entire course of the surface can be detected should, so that it is not sufficient, a local examination of the deviation of the Inclination to make an ideal surface. Rather, it must start out of at least two fixed points on which, for example, the curved rolled glass pane is supported, the inclination quasi stripes for stripes be determined so that the inclination multiplied by the width of the strip is a piece of a polygon. The next piece is then attached to this piece its "specific inclination", etc. This evaluation provides a large number of points of the course to be measured. For a graphic representation these points can then be displayed as a curve. For the The entire determined course can then determine the coordinates in space become. The evaluation can be done in one direction by means of a line camera a matrix camera in two mutually perpendicular directions.

Since the angular differences are measured, go advantageously in the case of measuring the flatness not in the calculation of the Such (absolute) height differences (e.g. in relation to a zero position of the device), which otherwise due to an oblique arrangement plan surface with conventional processes with sharp images work, would be determined based on the course of the image. Inaccuracies in the positioning of the glass 2 therefore adversely affect the Basically not measurement. Since the measurement including the evaluation in very short time (0.1 to 2 seconds), and a viewing period the surface 2 of a few milliseconds is sufficient for further processing, can easily move materials during their transportation. e.g. on conveyor belts or the like. Despite any vibrations can be measured without loss.

The light / dark values observed by the camera 5 are determined by an evaluation system (not shown) connected downstream of the camera evaluated. It is used as a reference for the evaluation of the distances of the Light grid 4 assumed that as a completely flat surface realistic reflection would have been reflected. The evaluation system determines 6 deviations from the light grid 4 in the mirror image and calculates from this the respective local angles by which the measured surface 2 of deviates from an ideal surface. It is also possible to do incremental Differences between adjacent strips as a basis for calculating the ripple the surface, which often has the optical properties of glass strong influenced and is therefore of great interest. In addition, the ripple can Providing an appropriately trained device on a rolled glass line can be used to control the rolling gas plant. Because the light at least in the first (only relevant here) approximation from the grid 4 parallel to the Disc 2 falls, it is advantageously not necessary to have a reference image record, and to base it on the evaluation, but it can by the known structure of the grid 4 (ie from the equidistant distances of the Grid 4) are assumed. The measurement error remains small and the Measuring time short.

2 and 3, the components from FIG. 1 are in one Measuring device shown in front view and side view. It's closed recognize that the camera 5 and the grid 4 at substantially the same height are arranged and enclose a relatively acute angle 8 of approximately 20 °. The Representation in Fig. 2 shows the grid 4 parallel, but slightly offset to the side Glass plate 2, i.e. at a small angle to the normal to surface 2 incident light beam (optical axis), and the camera 5 in the extension of the falling light reflected on the surface 2 (optical axis). In Fig. 3 can be seen that the section of the observed by the camera 5 Mirror image 6 depends on the aperture of the camera 5. That of the camera observed section is chosen so that it is 80 cm, so that 80 Light / dark pairs can be detected. If the number of streaks observed or decreases, is from the evaluation device to a corresponding Curvature of the surface like a lens (collecting or scattering) closed. Even within this curved contour Measure deviations in flatness quickly and accurately. It is possible in a few milliseconds and without the use of moving parts Surface topography of the test object 2 to be recorded, with measuring times of 0.1 to 2 seconds measuring accuracy between 0.1 to 3 µm (ripple) and Accuracy of <0.01 mm at a measuring length of approx. 800 to 1600 mm the course or flatness can be achieved.

In practice, the camera 5 is advantageously with the Raster unit 4 formed in a common assembly. The grid is 4 then slightly laterally parallel to an exactly opposite position staggered, and the camera 5 accordingly in the opposite Direction offset, so that preferably the same angle for both optical Axes. This allows a particularly favorable acute angle of realize a few degrees, for example 5 °, between the optical axes; the But the invention is also at opening angles of more than 5 ° and less than 90 ° practically applicable. It is also possible to place the camera in the grid integrate, and preferably both on a normal to surface 2 to arrange.

It is understood that the course of the device described is detected in one direction, namely that which is transverse to the direction of the Strip, that is, in the longitudinal direction (bend) of the rolled glass 2. in case of a essentially flat subjects, for example planning one Rolled glass section, on which a decor was printed, the Changes in inclination compared to an ideally planned course as a measure of the flatness can be used. Correspondingly, as above for the longitudinal course or planarity, the transverse course or planarity can also be determined are then subjected to the same measurement by the glass piece 2 , but with a relative rotation of 90 ° (either the system or the glass piece is arranged rotated by 90 °). For example, by two arrangements arranged one behind the other, rotated by 90 ° to one another 1 ensure that even in a continuous process, such as For example, in the production of rolled glass, information about the longitudinal and transverse flatness be determined. But it is also possible with a cross grid (Light / dark pairs in two, preferably perpendicular, directions) to project an image onto the reflecting surface of the glass piece 2, which is captured by a matrix camera. Then the course and Deviations from the course of an ideal subject in longitudinal and Transverse direction can be evaluated simultaneously, making measurements in particular easy to carry out on continuously or clocked conveyed surfaces 2 are.

It is thus possible to trace a three-dimensional sphere to be determined exactly, and also changes in gradient occurring therein To determine differentiation. If the surface 2 is flat, it can be very precisely Determine flatness; the surface 2 is curved (or it has a defined sphere on), the bending quality or smoothness can be determined very precisely.

As shown in Fig. 2 with dashed lines, there is one Peculiarity of the glass plate 2 in that it is double reflective, that is, that the grid 4 once on the top and once on the bottom of the Glass plate 2 is reflected. Because only the one at the top was not reflected Light beam (partially) reflected on the underside of the glass plate is the Intensity of the first reflection slightly stronger than the intensity of the second Reflection. So far the resulting mirror image 6 at an angle is observed, these two reflections overlap. The angle at the observation of the mirror image 6 is essentially due to the compact Size of the camera 5 and its aperture depends, and changes depending on Pixel; this angular offset is taken into account in the evaluation (the Path length change depending on the angle is determined by the Radiation ratio of the figure essentially canceled). It is possible, by a little unsharpness of the camera 5 the resultant of the overlay of the two mirror images in glass 2. The resultant of the of the Top of glass 2 reflected light intensity i1 and that of the bottom The light intensity i2 reflected by the glass is shown in FIG. 4. Because i1> i2 can the resulting signal can be easily separated so that the top and the Subpage with regard to their history (and other derived values like the Flatness, ripple, bending accuracy, height, thickness, etc.) can be evaluated separately.

Alternatively, it is possible to have a second camera 5 ′ behind the camera 5 to be arranged, whereby each of the two cameras on one of the mirror images (i.e. on the grid) is in focus and the acquired data is evaluated accordingly become. It is also possible to add further line cameras 5 "parallel to the camera 5 operate and therefore advantageous in parallel operation in a narrow zone many Record measured values and then process them further. Also a Scanning (scanning) with a line camera over an area possible.

The arrangement of a second camera 5 'at a different height than the camera 5, which preferably observes the same image section, still has another advantage. A change in the number of stripes observed can as already explained above, due to a lens shape of the surface his. However, the same phenomenon occurs if - for whatever reason always - the height of the observed surface 2 is changed and the observed image section contains more or less stripes. Due to the different heights of the cameras 5, 5 ', a height error can occur be easily corrected, and accordingly in the evaluation of the Course are taken into account.

Another alternative when evaluating double reflective materials, especially with tinted, e.g. green, glass are used. Becomes green glass (e.g. alternately) with red and with Illuminated green light from the grid, this is in very different Measure absorbed by the glass, so the intensity of that from the bottom reflected light differs significantly. From the different Intensities for different light wavelengths can be easily Conclusions about the position of the reflection on the top and bottom of the Pull glass. Alternatively, it is possible to observe a reflection of the mirror image 6 Use color camera, which makes the jumps in the detected intensities for Red, blue and green are easily separated and placed on the corresponding side of the Glases can be traced.

5 to 7, another embodiment a device according to the invention explained in more detail. The same reference numerals basically designate the same parts as in the previous ones Exemplary embodiments, from which it also follows that mutually substituting Parts are also interchangeable.

From the top view in Fig. 5 it can be seen that the device has two elongated grids 4a, 4b, which are directly next to each other are arranged. The two grids 4a, 4b both have stripes that are oblique (and not perpendicular to the previous embodiment) to it Longitudinal extension, in the present embodiment under one 45 ° angle. The slopes of the two grids 4a, 4b are at 90 ° to each other added.

Approximately in the middle of the two long sides facing away from each other Grid 4a, 4b two cameras 5a, 5b are arranged at approximately the same height observe the same image section 7. The observed object is a Endless belt e.g. made of flat glass or a partly reflective, partly transparent Plastic film. The cameras are a bit towards the frame 7 inclined. Each of the two cameras 5a, 5b observes in the present Embodiment the reflection of the grid further away from it. Alternatively, it would also be possible to place the cameras in the middle between the grids to arrange.

The arrow A in Fig. 7 indicates a direction of movement (current) of the Endless belt. X schematically illustrates a first type of error in FIG is typically effective in the direction of arrow A, i.e. that first at Passing the image section 7 a little further downstream (left in the figure) arranged area of the grid is recorded by the cameras 5a, 5b. The observed reflection then moves back again and further into the other Direction (opposite arrow A) before returning to the original area of the grid see is. This type of migration would be the same as in FIGS. 2 and 3 cannot be determined because the cameras always observe the same image have. In the present device, however, the camera 5a (left) captures first shift the (un-compressed / unstretched) raster image one side (e.g. to the right in the direction of arrow A), then in the reverse direction beyond the normal state to another Turn mark, and then another turn back to the original picture. The direction observed by the camera 5b (on the right in FIG. 5) is exactly reversed Hike. Thus one can use two line scan cameras and two grids Errors also determined in the plane transverse to the direction of extension of the grid become. For this purpose, the two grids 4a, 4b are provided with strips, which both to the conveying direction A of the film and to the mirror axis between the Line extension of the camera and the longitudinal extension of the grid at an angle are arranged.

An error of the type indicated by Y in FIG. 7 is caused by both Cameras reliably due to the compression or expansion of the observed Image, determined by a phase evaluating method as described, detected.

It goes without saying that surface courses or Flatness curves in which errors of type X and Y overlap, as in the reality is the case, reliable and inexpensive with two line scan cameras can be determined.

Claims (33)

1. A method for measuring the profile of a reflective surface of an object (2), comprising the steps
      aiming a defined pattern composed of at least two different light intensities onto the surface to be measured, wherein the pattern produces a mirror image (6) in the reflective surface;
      observing of at least one section of the surface by means of at least one camera (5), wherein the observed section comprises a section of the mirror image (6) of the pattern; and
      evaluating of the observed section based on the camera data,
   characterized in that the evaluation step comprises an integration of the angle profiles between two measurement points, from which it is possible to determine the geometric position of each point between the two measurement points wherein the profile of the surface is determined by appending calculated sections starting from a known point.
2. The method as claimed in claim 1, characterized in that the camera is set such that it is focused directly on the mirror image (6) of the pattern in the surface.
3. The method as claimed in claim 1, characterized in that the camera (5) observes the mirror image (6) via a mirror.
4. The method as claimed in one of claims 1 to 3, characterized in that the camera (5) is a line-scan camera or matrix camera.
5. The method as claimed in one of claims 1 to 4, characterized in that the pattern is composed of parallel, alternately light and dark strips.
6. The method as claimed in one of claims 1 to 4, characterized in that the pattern arranges light and dark squares like a checkerboard.
7. The method as claimed in one of claims 1 to 4, characterized in that the pattern is composed of mutually crossing lines of first brightness and of rectangles, preferably squares, enclosed by the lines and of a second brightness.
8. The method as claimed in one of claims 1 to 7, characterized in that the pattern is produced intermittently in time.
9. The method as claimed in one of claims 1 to 8, characterized in that the surface is moved relative to the light source (3) and camera (5).
10. The method as claimed in claim 9, characterized in that the step of observation of the mirror image (6) is fast in comparison with the rate of movement of the surface.
11. The method as claimed in one of claims 1 to 9, characterized in that a three-dimensional representation resulting from a single observation step is produced in the evaluation step.
12. The method as claimed in one of claims 1 to 11, characterized in that the surface to be measured is essentially planar, and the evaluation step comprises calculation of the local discrepancy from planarity.
13. The method as claimed in one of claims 1 to 12, characterized in that the evaluation step comprises the following steps
       calculation of the deflection of the mirror image (6) with respect to an ideal surface;
       determination of the inclination of the measured surface, based on the deflection.
14. The method as claimed in claim 13, characterized in that the evaluation step comprises the following steps
       integration of the inclination values in order to determine the profile of the surface over the observed section.
15. The method as claimed in claim 14, characterized in that the evaluation step comprises the following steps
       differentiation of the inclination values in order to determine the ripple of the surface over the observed section.
16. The method as claimed in one of claims 1 to 15, characterized in that at least one further camera observes the same image section as the at least one camera (5).
17. The method as claimed in one of claims 1 to 16, characterized in that two patterns are provided which run obliquely with respect to the mirror axis and whose mirror image is observed by in each case one line-scan camera (5a, 5b) in the same surface section (7) and that the lateral offset of the pattern observed by the two cameras (5a, 5b) is used for evaluating the observed sections (7), as a statement of the inclination change of the measured surface transversely with respect to the mirror axis.
18. The method as claimed in one of claims 1 to 17, characterized in that the observed object (2) is composed of a material which is at least partially transparent for light at specific wavelengths and which reflects this light on at least one further surface, arranged behind the surface.
19. An apparatus for determining the profile of the reflective surface of an object, comprising
       means (3,4) for producing a light pattern;
       at least one camera (5) for observing at least one section of the surface, and
       means for evaluating of the observed section based on the camera data,
    characterized in that the camera (5) is set focused on a mirror image (6) of the light pattern in the surface,
    that the means for evaluating comprise in use integrating of the angle profiles between two measurement points, and
    that the means for evaluating comprise in use calculating the geometric position of each point between the two measurement points, starting from a known point to which a corresponding number of evaluated sections are appended.
20. The apparatus as claimed in claim 19, characterized in that the means for producing a light pattern comprise a light source (3) and a grid (4) arranged between the light source and the surface.
21. The apparatus as claimed in claim 20, characterized in that the means for producing a light pattern comprise a light wall composed of a matrix of illumination sources, preferably LEDs, which can be driven individually.
22. The apparatus as claimed in one of claims 19 to 21, characterized in that the optical axis between the light pattern and the mirror image (6) on the one hand, and between the mirror image (6) and the camera (5) on the other hand, together include an angle (8) of less than 90°.
23. The apparatus as claimed in one of claims 19 to 22, characterized in that the light pattern has a sharply bounded structure of at least two different light intensities which are arranged regularly and alternating.
24. The apparatus as claimed in one of claims 19 to 23, characterized in that the camera (5) is a matrix camera or a line-scan camera, and that a line of the camera (5) and the means for producing the light pattern extend parallel to one another.
25. The apparatus as claimed in one of claims 19 to 24, characterized in that the means for producing the light pattern (3, 4) and the camera (5) are arranged in a housing at a short distance from one another or integrated in one another above the surface (2).
26. The apparatus as claimed in one of claims 19 to 25, characterized in that at least one defined mounting point is provided for the object to be measured.
27. The apparatus as claimed in one of claims 19 to 26, characterized in that the surface (2) is composed of glass.
28. The apparatus as claimed in claim 27, characterized in that the glass is in the form of an endless strip, and that means are provided in order to feed the glass through the apparatus.
29. The apparatus as claimed in one of claims 19 to 28, characterized in that a parabolic mirror is arranged in such a manner that the mirror image (6) falls on the camera via the parabolic mirror.
30. The apparatus as claimed in one of claims 19 to 29, characterized in that two cameras are arranged, and are pointed essentially at the same section of the mirror image (6).
31. The apparatus as claimed in one of claims 19 to 30, characterized in that the camera can be pivoted or can be moved in order that sections which are adjacent to the section can also be scanned.
32. The apparatus as claimed in one of claims 19 to 31, characterized in that the means for producing a light (4), the camera (5) and the surface (2) are arranged on essentially parallel planes.
33. The apparatus as claimed in one of claims 19 to 32, characterized in that two means for producing a light pattern are provided, which are at an angle with respect to one another and with respect to the mirror axis with in each case one associated camera, and that the two cameras observe the mirror image of in each case one pattern at the same reflection point on the surface.

Images